Newborns• Cry Melody Is Shaped by Their Native Language

Birgit Mampe,1 Angela D. Friederici,2 Anne Christophe,3 contours (exempliﬁed in Figure 2). These results show and Kathleen Wermke1,* a tendency for infants to utter melody contours similar to those 1Center for Prespeech Development and Developmental perceived prenatally. A signiﬁcant difference was also found

Disorders, Department of Orthodontics, for the intensity maxima of melody arcs [tnorm(Imax): mean University of Wu¨rzburg, 97070 Wu¨rzburg, Germany 0.59 6 0.12 versus 0.47 6 0.12 for French group versus 2Max-Planck-Institute for Human Cognitive and Brain German group; Mann-Whitney test, p < 0.001]. The difference Sciences, 04103 Leipzig, Germany in the median values of tnorm(Imax)—0.61 s versus 0.45 s for 3Laboratoire de Sciences Cognitives et Psycholinguistique, French versus German—are also displayed in Figure 1. Ecole Normale Supe´ rieure/CNRS, 75005 Paris, France In addition, melody and intensity were significantly corre- lated in both groups (Spearman rho 0.45, p < 0.05 and 0.69, p < 0.01 for the French group and German group, respectively). Summary However, despite the robust correlation between melody and intensity, there is some evidence that they are controlled by Human fetuses are able to memorize auditory stimuli from separate neurophysiological mechanisms [19–21]. Indeed, the external world by the last trimester of pregnancy, with several cries exhibited independent melody and intensity a particular sensitivity to melody contour in both music contours. and language [1–3]. Newborns prefer their mother’s voice over other voices [4–8] and perceive the emotional content Discussion of messages conveyed via intonation contours in maternal speech (‘‘motherese’’) [9]. Their perceptual preference for Prosodic features such as melody, intensity, and rhythm are the surrounding language [10–12] and their ability to distin- essential for an infant acquiring language [22]. There is guish between prosodically different languages [13–15] and compelling evidence that infants are sensitive to prosodic pitch changes [16] are based on prosodic information, features of their native language long before speech-like primarily melody. Adult-like processing of pitch intervals babbling sounds are uttered or first words are produced allows newborns to appreciate musical melodies and [22, 23]. Indeed, auditory learning starts as early as the third emotional and linguistic prosody [17]. Although prenatal trimester of gestation [24, 25], and prosodic features are well exposure to native-language prosody influences newborns’ preserved across the abdominal barrier, whereas phonetic perception, the surrounding language affects sound produc- aspects of speech are disrupted, making prosodic character- tion apparently much later [18]. Here, we analyzed the crying istics very salient for the human fetus [26]. In newborns, traces patterns of 30 French and 30 German newborns with respect of early auditory learning processes are reflected in perceptual to their melody and intensity contours. The French group preferences for melodies to which they were exposed prena- preferentially produced cries with a rising melody contour, tally [1, 10, 14, 27, 28]. whereas the German group preferentially produced falling The present study is different from former investigations in contours. The data show an influence of the surrounding that it focuses on a possible influence of the surrounding speech prosody on newborns’ cry melody, possibly via language on newborns’ sound production instead of investi- vocal learning based on biological predispositions. gating perceptual preferences for the native language. This influence was investigated by analyzing melody contours of Results newborns’ crying. The observed melody contours of French and German Cries of 60 healthy newborns, 30 born into French and 30 born newborns’ crying show that they not only have memorized into German monolingual families, were analyzed. Melody in the main intonation patterns of their respective surrounding neonates’ cries is characterized by single rising-and-then- language but are also able to reproduce these patterns in their falling arcs. These melody arcs were analyzed by determining own production. Newborns produced significantly more often the relative (normalized) time at which the maximum pitch those melody types and intensity contours that were prosodically typical for their native languages: French newborns pref- (F0max) was reached [tnorm(F0max)] (see ‘‘Melody Contour Anal- ysis’’ in Experimental Procedures). Intensity contour analyses erentially produced rising (low to high) contours, whereas were performed in a corresponding manner for each cry. German newborns preferentially produced falling (high to As shown in Figure 1, a marked difference in the median low) contours (for both melody and intensity contours). These patterns are consistent with the intonation patterns values of tnorm(F0max) points to group-specific preferences for produced melody contours (French group, 0.60 s; German observed in both of these languages. In French, intonation is characterized by a pitch rise toward the end of several kinds group, 0.45 s). The arithmetic means of tnorm(F0max) were significantly different in French (0.58 6 0.13 s) and German of prosodic units (words, intermediate prosodic phrases), (0.44 6 0.15 s) newborns (Mann-Whitney test, p < 0.0001). except for the very last unit of an utterance, which presents Whereas French newborns preferred to produce rising melody a falling contour (see, e.g., [29, 30]). This is a crucial difference contours, German newborns more often produced falling from German intonation, which typically exhibits a falling melody contour, e.g., from the accented high-tone syllable to the end of the intonational phrase [31]. This difference between *Correspondence: [email protected] French and other languages has already been observed to Cry Melody of French and German Newborns 1995

Figure 1. Box-Plot Diagram of the Values tnorm(F0max) and tnorm(Imax) Distribution of all observed melody and intensity contours in German and French newborns’ crying, displayed as box plots of the 25th to 75th percen- tile, with the solid vertical line inside each box representing the median and the bars outside each box representing the minimum and maximum values. The dashed vertical line represents a symmetric melody arc. The data indicate a preference for either rising (French group) or falling (German group) melodies. have an impact on the sound production of 7- to 18-month-old infants: French infants have been found to produce more rising melody contours than English and Japanese infants [32, 33]. The newborns examined in the present study probably learned these characteristics of their mother tongue by listening to it prenatally (although we cannot completely Figure 2. Time Waveform and Narrow-Band Spectrograms of a Typical exclude early postnatal learning during the first 2–5 days of French Cry and a Typical German Cry life). Language-specific preferences for final versus initial stress patterns have already been reported in perception for French and German infants as young as 4 months of age via follow a falling pattern, a simple physiological consequence of neurophysiological techniques [34]. The present cry produc- the rapidly declining subglottal pressure during expiratory tion data show an even earlier impact of native language, phonation. The present data show that German and French because neonates’ cries are already tuned toward their native infants produce different types of cries, even though they language. share the same physiology. In particular, the fact that the The specific perceptual abilities of human fetuses and young French newborns produce ‘‘nonphysiological’’ rising patterns infants for melody properties evolved over several million years supports former findings demonstrating that human newborns’ of vocal and auditory communication and (more recently) cry melody patterns are already a product of a well-coordi- spoken language [35]. Thus, rather than being specific to nated respiratory-laryngeal activity under the control of neuro- speech, most of the precocious perceptual performances of physiological mechanisms [19, 20]. human infants have deep roots in a phylogenetically older Apes’ vocalizations (e.g., bonobo sounds) are described as primate auditory perceptual system. There are also obvious exhibiting a strict correlation between F0 variations and inten- acoustic similarities between nonhuman primate calls and sity, suggesting a close association between F0, intensity, human infant cries (cf. review in [36, 37]). However, in contrast and subglottal pressure [41]. Many of the laryngeal muscle func- to nonhuman primates, human infants develop spoken tions for swallowing, respiration, and vocalization are controlled language quickly and seemingly without effort. In spite of by subcortical regions in nonhuman primates [42], whereas many similarities, human infants and nonhuman primates differ intentional control of breathing and crying in newborns origi- with respect to language-relevant perceptive capacities (cf. nates in the cerebral cortex [36, 43]. The muscles of the larynx [38]) as well as early productive performances. function as a part of the respiratory system before birth. Like Thus, two aspects of the present data suggest that human other respiratory muscles, they undergo considerable use prior infants’ melody production is based on a well-coordinated to birth [44]. For reproducing the melody contours perceived respiratory-laryngeal activity, in contradiction to older studies and stored prenatally, a coordinated action of melody and inten- that argued that cry melody was strictly constrained by the sity may simply be a very economic and thus the easiest way to respiratory cycle (e.g., [39, 40]). First, newborns seem capable achieve the contour target, even though infants will manipulate of an independent control of fundamental frequency and inten- these parameters independently in the process of learning to sity, as suggested by the observed cases of cries where speak. melody and intensity contours are decorrelated (see also The present cry production data show an extremely early [19]). Second, and more importantly, if newborns’ cries were impact of native language. Thus far, a capability for vocal constrained by the respiratory cycle, then they should always imitation had only been demonstrated from 12 weeks onward. Current Biology Vol 19 No 23 1996

Listening to an adult speaker producing vowels, infants re- sponded with utterances that perceptually matched the vowels presented to them [45]. To do this, infants must be capable of moving their articulators in order to reach a specific auditory target. Anatomical and functional constraints of the immature vocal tract mechanisms do not allow for the imitation of articulated speech sounds before about 3 months. Imitation of melody contour, in contrast, is merely predicated upon well- coordinated respiratory-laryngeal mechanisms and is not constrained by articulatory immaturity. Newborns are probably highly motivated to imitate their mother’s behavior in order to attract her and hence to foster bonding [46, 47]. Because melody contour may be the only aspect of their mother’s speech that newborns are able to imitate, this might explain Figure 3. Diagrammed Cry Melody as Time Function of Fundamental Frequency F0 with Time-Normalized Duration why we found melody contour imitation at that early age. Hence, our data support the existence of imitation in newborns caused by strong nonlinearities in the restoring forces resulting from an by fulfilling all the necessary prerequisites postulated by Jones extremely large amplitude-to-length ratio of the vocal folds in newborns [48]. Whether human newborns’ preference for speech is [53]. From a perspective of nonlinear dynamics, voiceless (noisy) segments innate [49] or acquired [26], the observed performances are of newborn cries can be interpreted as low-dimensional chaos [53, 54]. based on biological predispositions, particularly for melody Voiceless cries and cries containing broad regions of phonatory noise in perception and production [47, 50]. their frequency spectra could not be used for the applied methodology because fundamental frequency (melody) cannot be reliably determined in Recent findings indicate a systematic melody development those signals. from simple to complex patterns beginning at birth and For the final melody analyses, 1254 voiced cries were used (French group, demonstrate a strong developmental continuity from crying mean number of cries per neonate: 21, range 3–54; German group, mean via cooing and babbling toward speech (e.g., [20, 21, 50–52]). number of cries per neonate: 18, range 10–38). The large interindividual vari- The significant finding of this study is that newborns, in their ability in number of cries per session was due to the fact that we strictly own sound production, already reproduce some of the avoided eliciting crying (through stimulation). prosodic properties of the specific language that they were exposed to prenatally. Melody Contour Analysis Only simple cries containing single rising-and-then-falling melody arcs were analyzed. These cry types were selected because they predominate in the Experimental Procedures crying of healthy newborns. These melody arcs can be assigned to four basic melody types: (1) quickly rising and slowly decreasing melody: Participants left-accentuated type—‘‘falling contour’’; (2) slowly rising and quickly Cries of 30 French (11 female, 19 male; mean age 3.1 days, range 2–5 days) decreasing melody: right-accentuated type—‘‘rising contour’’; (3) symmet- and 30 German (15 female, 15 male; mean age 3.8 days, range 3–5 days) rical rising-and-then-falling melody: symmetric type; and (4) a relatively newborns were analyzed. All subjects were healthy, full-term newborns stable fundamental frequency with a rising or falling trend: plateau type with normal hearing from a strictly monolingual (French or German) family [20, 50]. These melody arcs were analyzed as follows (see Figure 3): after background. German infants had a mean gestational age of 39.5 weeks; normalizing arc duration to 1 s, the normalized time [tnorm(F0max)] corre- French newborns had a mean gestational age of 39.6 weeks. The studies sponding to the maximum pitch (F0max) of each melody arc was determined. were performed with the approval of the ethics committees of Charite´ Berlin tnorm(F0max) values < 0.5 s represent ‘‘falling contours’’; tnorm(F0max) values > and Cochin Hospital (Paris). All participating families followed the institu- 0.5 s represent ‘‘rising contours’’ (Figure 1). Intensity contour analyses were tional consent procedures in German or French. performed in a corresponding manner for each cry. Newborns of both groups generated all four basic melody types typical at Cry Recordings that age. This observation reflects a general aptitude for generating melo- Recordings of the French newborns were made at Port-Royal Maternity of dies with varying contours and explains the observed partial data overlap Cochin Hospital (Paris). For the German newborns, existing digital sound in Figure 1. files of cries that were recorded as part of the German Language Develop- ment Study (http://glad-study.cbs.mpg.de) were used. Cry recordings were Acknowledgments made during spontaneous, normal mother-child interactions (while changing diapers, before feeding, or while calming the spontaneously fussy This work was supported by the European Union (EC12778/NEST-CALACEI baby) with a TASCAM DAT recorder (DA-P1, serial number 880096) and an Project) and a grant to B.M. from the FAZIT-Foundation. Earthworks microphone (TC20, serial number 7642C). All recordings were made in pain-free situations (excluding, e.g., pain cries in response to heel lancing for blood sampling). During recording, the distance between Received: August 20, 2009 the microphone and the newborn’s mouth was about 15 cm. Individual Revised: September 27, 2009 recording time varied between 3 and 10 min, depending on the spontaneous Accepted: September 28, 2009 crying behavior of the neonate. A cry was defined as the vocal output occur- Published online: November 5, 2009 ring on a single expiration. In total, 2500 cries were recorded. References Cry Analysis Spectral analysis and high-resolution melody (fundamental frequency [F0] 1. DeCasper, A.J., and Spence, M.J. (1986). Prenatal maternal speech contour) computations were carried out with Computerized Speech Labora- influences newborns’ perception of speech sounds. Infant Behav. tory CSL 4400/MDVP (Kay Elemetrics Corp.) together with postprocessing Dev. 9, 133–150. for interactive removal of high-frequency modulation noise and artifacts 2. Kisilevsky, S., Hains, S.M., Jacquet, A.-Y., Granier-Deferre, C., and (Cry-Data-Analysis-Program [CDAP]; see [19] for details). Frequency spec- Lecanuet, J.P. (2004). Maturation of fetal responses to music. Dev. trograms of all cries were produced in order to identify voiced, harmonic Sci. 7, 550–559. cries and to exclude noisy, voiceless cries. Depending on physical state 3. Granier-Deferre, C., Bassereau, S., Jacquet, A.Y., and Lecanuet, J.P. and the course of postnatal adaptation, newborns’ crying may contain irreg- (1998). Fetal and neonatal cardiac orienting response to music in quiet ularities such as subharmonics or phonatory noise. Such phenomena are sleep. Dev. Psychobiol. 33, 372. Cry Melody of French and German Newborns 1997

4. Querleu, D., Lefebvre, C., Titran, M., Renard, X., Morillion, M., and 32. Halle´ , P.A., de Boysson-Bardies, B., and Vihman, M.M. (1991). Begin- Crepin, G. (1984). [Reaction of the newborn infant less than 2 hours after nings of prosodic organization: Intonation and duration patterns of birth to the maternal voice]. J. Gynecol. Obstet. Biol. Reprod. (Paris) 13, disyllables produced by Japanese and French infants. Lang. Speech 125–134. 34, 299–318. 5. DeCasper, A.J., and Fifer, W.P. (1980). Of human bonding: Newborns 33. Levitt, A.G., and Wang, Q. (1991). Evidence for language-specific prefer their mothers’ voices. Science 208, 1174–1176. rhythmic influences in the reduplicative babbling of French- and 6. Ockleford, E.M., Vince, M.A., Layton, C., and Reader, M.R. (1988). English-learning infants. Lang. Speech 34, 235–249. Responses of neonates to parents’ and others’ voices. Early Hum. 34. Friederici, A.D., Friedrich, M., and Christophe, A. (2007). Brain Dev. 18, 27–36. responses in 4-month-old infants are already language specific. Curr. 7. Damstra-Wijmenga, S.M. (1991). The memory of the new-born baby. Biol. 17, 1208–1211. Midwives Chron. 104, 66–69. 35. Mithen, S. (2006). The Singing Neanderthals: The Origins of Music, 8. Hepper, P.G., Scott, D., and Shahidullah, S. (1993). Newborn and fetal Language, Mind and Body (London: Phoenix). response to maternal voice. J. Reprod. Infant Psychol. 11, 147–153. 36. Newman, J.D. (2007). Neural circuits underlying crying and cry respond- 9. Fernald, A., and Simon, T. (1984). Expanded intonation contours in ing in mammals. Behav. Brain Res. 182, 155–165. mothers’ speech to newborns. Dev. Psychol. 20, 104–113. 37. Masataka, N. (2008). The Onset of Language (Cambridge: Cambridge 10. Mehler, J., Jusczyk, P.W., Lambertz, G., Halsted, N., Bertoncini, J., and University Press). Amiel-Tison, C. (1988). A precursor of language acquisition in young 38. Pinker, S., and Jackendoff, R. (2005). The faculty of language: What’s infants. Cognition 29, 143–178. special about it? Cognition 95, 201–236. 11. Moon, C., Cooper, R.P., and Fifer, W.P. (1993). Two-day-olds prefer 39. Lieberman, P. (1967). Intonation, Perception, and Language (Cam- their native language. Infant Behav. Dev. 16, 495–500. bridge, MA: MIT Press). 12. Mehler, J., and Dupoux, E. (1994). What Infants Know (Oxford: Basil 40. Lieberman, P. (1985). The Physiology of Cry and Speech in Relation to Blackwell). Linguistic Behavior. In Infant Crying: Theoretical and Research 13. Mehler, J., and Christophe, A. (1995). Maturation and learning of Perspectives, B.M. Lester and Z.C.F. Boukydis, eds. (New York: Plenum language in the first year of life. In The Cognitive Neurosciences: A Press), pp. 29–57. Handbook for the Field, M.S. Gazzaniga, ed. (Cambridge, MA.: MIT 41. Demolin, D., and Delvaux, V. (2006). A comparison of the articulatory Press), pp. 943–954. parameters involved in the production of sound of bonobos and modern 14. Nazzi, T., Floccia, C., and Bertoncini, J. (1998). Discrimination of pitch humans. In Proceedings of the Sixth International Conference on the contours by neonates. Infant Behav. Dev. 21, 779–784. Evolution of Language, A. Cangelosi, K. Smith, and A.D.M. Smith, eds. 15. Ramus, F., Hauser, M.D., Miller, C., Morris, D., and Mehler, J. (2000). (Rome: World Scientific Publishing Co.), pp. 67–74. Language discrimination by human newborns and by cotton-top 42. Simonyan, K., and Ju¨rgens, U. (2003). Efferent subcortical projections of tamarin monkeys. Science 288, 349–351. the laryngeal motorcortex in the rhesus monkey. Brain Res. 974, 43–59. 16. Carral, V., Huotilainen, M., Ruusuvirta, T., Fellman, V., Na¨ a¨ ta¨ nen, R., and 43. Darnall, R.A., Ariagno, R.L., and Kinney, H.C. (2006). The late preterm Escera, C. (2005). A kind of auditory ‘primitive intelligence’ already infant and the control of breathing, sleep, and brainstem development: present at birth. Eur. J. Neurosci. 21, 3201–3204. A review. Clin. Perinatol. 33, 883–914. 17. Stefanics, G., Ha´ den, G.P., Sziller, I., Bala´ zs, L., Beke, A., and Winkler, I. 44. Harding, R. (1984). Function of the larynx in the fetus and newborn. (2009). Newborn infants process pitch intervals. Clin. Neurophysiol. 120, Annu. Rev. Physiol. 46, 645–659. 304–308. 45. Kuhl, P.K., and Meltzoff, A.N. (1996). Infant vocalizations in response to 18. Whalen, D.H., Levitt, A.G., and Wang, Q. (1991). Intonational differences speech: Vocal imitation and developmental change. J. Acoust. Soc. Am. between the reduplicative babbling of French- and English-learning 100, 2425–2438. infants. J. Child Lang. 18, 501–516. 46. Meltzoff, A.N., and Moore, M.K. (1983). Newborn infants imitate adult 19. Wermke, K., Mende, W., Manfredi, C., and Bruscaglioni, P. (2002). facial gestures. Child Dev. 54, 702–709. Developmental aspects of infant’s cry melody and formants. Med. 47. Falk, D. (2009). Finding Our Tongues (New York: Basic Books). Eng. Phys. 24, 501–514. 48. Jones, S.S. (2009). The development of imitation in infancy. Philos. 20. Wermke, K. (2002). [Investigation of the melody development in cries of Trans. R. Soc. Lond. B Biol. Sci. 364, 2325–2335. monozygotic twins in the first 5 months of life]. Habilitationsschrift 49. Vouloumanos, A., and Werker, J.F. (2007). Listening to language at birth: (postdoctoral thesis), Humboldt-Universita¨ t zu Berlin, Berlin. http:// Evidence for a bias for speech in neonates. Dev. Sci. 10, 159–164. edoc.hu-berlin.de. 50. Wermke, K., and Mende, W. (2009). Musical elements in human infants’ cries: In the beginning is the melody. In Musicae Scientiae, Special Issue 21. Wermke, K., and Mende, W. (1994). Ontogenetic development of infant on Music and Evolution, O. Vitouch and O. Ladinig, eds. (Brussels: cry and non-cry vocalization as early stages of speech abilities. In Presses Universitaires de Bruxelles), pp. 151–173. Proceedings of the Third Congress of the International Clinical 51. Wermke, K., and Friederici, A.D. (2004). Developmental changes of Phonetics and Linguistics Association, R. Aulanko and A.M. Korpi- infant cries – The evolution of complex vocalizations. Behav. Brain jaakko-Huuhka, eds. (Helsinki, Finland: University of Helsinki), pp. Sci. 27, 474–475. 181–189. 52. Hsu, H.C., Fogel, A., and Cooper, R.B. (2000). Infant vocal development 22. Vihman, M.M. (1996). Phonological Development: The Origins of during the first 6 months: Speech quality and melodic complexity. Infant Language in the Child (Oxford: Blackwell Publishers). Child Dev. 9, 1–16. 23. Boysson-Bardies, B. (1999). How Language Comes to Children: From 53. Titze, I.R., Baken, R.J., and Herzel, H. (1993). Evidence of chaos in vocal Birth to Two Years (Cambridge, MA: MIT Press). fold vibration. In Vocal Fold Physiology, I.R. Titze, ed. (San Diego, CA: 24. Hepper, P.G. (1997). Memory in utero? Dev. Med. Child Neurol. 39, 343– Singular Publication Group), pp. 143–188. 346. 54. Mende, W., Herzel, H., and Wermke, K. (1990). Bifurcations and chaos in 25. Shahidullah, S., and Hepper, P.G. (1994). Frequency discrimination by newborn infant cries. Phys. Lett. A 145, 418–424. the fetus. Early Hum. Dev. 36, 13–26. 26. Rosen, S., and Iverson, P. (2007). Constructing adequate non-speech analogues: What is special about speech anyway? Dev. Sci. 10, 165– 168. 27. James, D.K., Spencer, C.J., and Stepsis, B.W. (2002). Fetal learning: A prospective randomized controlled study. Ultrasound Obstet. Gynecol. 20, 431–438. 28. Hepper, P.G. (1988). Fetal ‘‘soap’’ addiction. Lancet 1, 1347–1348. 29. Delattre, P. (1961). [The intonation model of Simone de Beauvoir: A declarative comparative study on intonation]. French Review 35, 59–67. 30. Welby, P. (2006). French intonational structure: Evidence from tonal alignment. J. Phon. 34, 343–371. 31. Wiese, R. (1996). The Phonology of German (Oxford: Clarendon Press).